- Email: [email protected]
Increased familiarity with eating a food to fullness underlies increased expected satiety
Appetite 61 (2013) 13–18
Contents lists available at SciVerse ScienceDirect
Appetite journal homepage: www.elsevier.com/locate/appet
Research report
Increased familiarity with eating a food to fullness underlies increased expected satiety q Michael A. Irvine a,c,⇑, Jeffrey M. Brunstrom a, Philip Gee b, Peter J. Rogers a a
School of Experimental Psychology, University of Bristol, 12a Priory Road, Bristol BS8 1TU, England, United Kingdom School of Psychology, University of Plymouth, Drake Circus, Plymouth PL4 8AA, England, United Kingdom c Department of Psychiatry, University of Cambridge, Addenbrooke’s Hospital site, Level E4, P.O. Box 189, Cambridge CB2 0QQ, United Kingdom b
a r t i c l e
i n f o
Article history: Received 29 March 2012 Received in revised form 19 September 2012 Accepted 11 October 2012 Available online 22 October 2012 Keywords: Expected satiety Expected satiation Palatability Energy density Associative learning Flavour–nutrient learning Learned satiety
a b s t r a c t Expected satiety informs self-selected portion sizes and thereby influences energy intake. At present the extent to which these beliefs are learned remains unclear. In an initial study the proposition that familiarity influences expected satiety was explored. Self-report measures of familiarity, along with other measures such as degree of liking, were collected for wine gums and milk chocolate, together with expected satiety estimates obtained using a psychophysical task. Familiarity was indeed significantly correlated with expected satiety, but only in respect of frequency of having eaten the food to fullness. In a second experiment a significant increase in expected satiety was observed after eating a large portion of wine gums at a subsequent test session. Together, these findings indicate that expected satiety changes in response to increased familiarity of eating a food to satiety. Ó 2012 Elsevier Ltd. All rights reserved.
Introduction Prospective judgements relating to the ‘‘fillingness’’ of foods are referred to as decisions based on ‘‘expected satiety’’. Somewhat surprisingly, given their potential import to energy intake, these judgements have historically been given little consideration. As Brunstrom and Shakeshaft (2008) report, however, satiety expectations are an important determinant of self-selected portion size (kcal), and as such they would appear to be very relevant to the area of weight control. This is because portion size has been shown to be an excellent predictor of amount of food consumed (Jeffery et al., 2007; Rolls, Roe, Kral, Meengs, & Wall, 2004a; Rolls, Roe, & Meengs, 2006a, 2006b, 2007; Rolls, Roe, Meengs, & Wall, 2004b; Wansink, Painter, & North, 2005), where self-selected portions are often eaten in their entirety (Krassner, Brownell, & Stunkard, 1979; Lebow, Chipperfield, & Magnusson, 1985; Wansink & Cheney, 2005). Indeed, and even after controlling for volume, Brunstrom, Collingwood, and Rogers (2010) found a close correspondence between expected satiation and self-selected portion size.
q Acknowledgments: This work was supported by Cadbury plc, a member of Kraft FoodsTM and Great Western Research. A preliminary report of these experiments was presented at the 8th Sensory Science Symposium, Florence, Italy (2009). ⇑ Corresponding author. E-mail address: [email protected] (M.A. Irvine).
0195-6663/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.appet.2012.10.011
Presumably, eating experiences shape beliefs about the expected ‘‘fillingness’’ of foods. Therefore expected satiety should correspond with the actual degree of satiety occurring after ingestion. Supporting this, Brunstrom, Shakeshaft, and Scott-Samuel (2008) reports a highly significant correlation between the expected satiety scores of participants in their study with the actual satiety scores derived in a study by Holt, Brand-Miller, and Petocz (1995). In addition these authors found that familiar foods are expected to be more satiating. One possibility is that different aspects of familiarity – for example familiarity with eating a food per se (that is, in any caloric denomination) versus familiarity of eating it in sufficient quantities so as to confer satiety – might be differentially important in informing satiety expectations. Brunstrom et al. (2008) suggests that certain foods may be expected to confer relatively less satiety (kcal for kcal) because they are not often eaten to satiety, thereby limiting the opportunity for learned satiety to occur. Furthermore, Rogers and Smit (2000) allude to the possibility that snack foods may be conceptualised as ‘‘moreish’’ due to the fact societal norms dictate that such foods are relatively rarely eaten in amounts equivalent to staple foods (i.e., in amounts that generate significant to strong feelings of fullness). In light of the above discussion, the aim of the present paper was to further explore the relationship between expected satiety and familiarity, and in particular familiarity of eating to fullness. In Experiment 1 we sought to establish whether expected satiety
14
M.A. Irvine et al. / Appetite 61 (2013) 13–18
is correlated with self-reported ratings of familiarity for milk chocolate and wine gums (a chewy, pastille-type sweet/candy, which is generally eaten less frequently than chocolate). Familiarity was indexed by dissociable measures specifically pertaining to either familiarity of eating a food per se, or familiarity of eating it to fullness. In Experiment 2 an eating component was introduced: ratings of expected satiety for wine gums (the relatively novel food) were recorded before and after an eating episode. We reasoned that a difference between pre- and post-eating episode ratings would provide evidence of a direct influence of (increased) familiarity on expected satiety beliefs. Method – Experiment 1 Participants The participants were 7 males and 39 females. They were drawn from the undergraduate population of the University of Bristol. Their mean age was 19.4 [S.D. = 5.8] years. No specific exclusion criteria were applied. They participated as part of their course requirement, or were offered £7 Sterling in remuneration for their assistance with the study. Most participants were of normal body weight. The University of Bristol Faculty of Science Human Research Ethics Committee gave approval for the protocol. Design Images of seven different foods were presented in 260-kilocalorie (kcal) portions, as follows: pasta and sauce (‘‘Egg Penne’’ pasta, Sainsbury’s UK, Sainsbury’s Supermarkets Ltd., Holborn, London, UK); ‘‘Sun-dried stir in tomato sauce’’ (Dolmio, Masterfoods, Melton Mowbray, Leicestershire, UK); pizza (‘‘Deeply Delicious Margherita’’, Goodfellas, Northern Foods Frozen, Salford Quays, Manchester, UK); Dairy Milk, Crunchie, and Maynard’s Wine Gums (all three by Cadbury plc, Bournville, Birmingham, UK); Peanut M&Ms (Mars, Mars UK Ltd., Slough, Berkshire, UK); and banana. Dairy Milk is a solid milk chocolate bar; Crunchie is a honeycomb-centred, chocolate-covered bar; and Wine Gums are chewy, firm, fruit-flavoured pastille-type sweets/candy. These foods were used because previous studies have shown that they vary in familiarity, liking, and the typical context in which they are eaten (Irvine et al., unpublished data). For the present study data pertaining to these measures are given in Table 2 (see Results section). The macronutrient composition and energy density of these foods are shown in Table 1. Measures Psychophysical task An adapted version of the psychophysical methodology first introduced by Brunstrom et al. (2008) was employed to assess expected satiety. This methodology is based on a ‘‘method of con-
Table 1 Macronutrient composition and energy density of the seven foods. Values for carbohydrate, protein, fat, fibre and kcals are per 100 g. Foods are ordered from least to most energy dense. Food
Carbohydrate
Protein
Fat
Fibre
kcal
Banana Pasta & sauce Wine gums Pizza Crunchie M&Ms Dairy milk
23.2 21.4 75 40.2 69.9 46.4 56.7
1.2 5.3 6.1 16.5 3.6 9.6 7.5
0.3 5.3 0.2 13.2 19.1 26.2 29.8
1.1 3.1 0 4 1.3 2.8 0.7
95 153 325 346 465 515 525
stant stimuli’’ (MOCS). The MOCS is well suited to the measurement of expected satiety across different foods. In comparison to methods where estimates are based on a single decision, the MOCS is used to derive estimates based upon many decisions, which promotes greater accuracy. Estimates of expected satiety were obtained for the same seven foods described above. Five of the foods were designated ‘‘comparison foods’’. The other two, milk chocolate and wine gums, were the ‘‘standard foods’’, and were always depicted in 255 kcal (milk chocolate) and 260 kcal (wine gums) portions. These caloric denominations were used because there are 255 kcals in a standard 49 g bar of Dairy Milk™, and there are 260 kcals in 13 wine gums, which gives the closest match to the Dairy Milk bar without having to cut up any wine gums. The assignment of a food as a comparison or a standard is somewhat arbitrary, in that the MOCS procedure can work perfectly well in any configuration (see for example Brunstrom et al., 2008). As such there is no basis on which to predict that results will depend upon which foods were designated to which category. At any given time the computer screen displayed a photograph of one of the two standard foods on the left and one of the five comparison foods on the right, arranged on a 255-mm diameter white plate. The images were of a high quality, and were taken using a camera mounted directly overhead. For each comparison food there were between 40 and 70 photographs taken, of different portion sizes varying in 20 kcal increments. As such, the maximum displayed portion sizes for the comparison foods varied between 800 kcal and 1400 kcal. This difference was due to the fact some of the less energy-dense foods could not be accommodated on the plate in very large portions (by weight). During the task the participants pressed the left cursor key if they thought the food portion on the left would make them feel the most full, or the right cursor key to indicate the portion on the right. After each key press the next two portions of food would appear. On each trial the amount and type of comparison food changed. Each standard food was paired with each comparison food, in various caloric denominations, on multiple occasions. After a sufficient number of trials the probability that the standard will be selected over the range of comparison values can be plotted: probit analysis can be used to fit a sigmoid function, from which a point of subjective equality (PSE) can be extrapolated. As noted by Brunstrom et al. (2008), this value is important because it ‘‘indicates the amount of the comparison (kcal) that is expected to be equally as filling as the standard’’. Thus, and by making a set of systematic comparisons between a common standard/s and a range of comparison foods, precise expected satiety scores can be derived and compared across a range of foods. In total participants responded to 540 trials (54 pairings of each of the five comparison foods with each of the two standard foods). Attitudes to foods task In this task participants responded to 11 different statements, which were to be correlated with estimates of expected satiety. To reiterate, the rationale was to assess the relationship between expected satiety and attitudes towards/level of familiarity with the foods. The statements were as follows: 1. 2. 3. 4. 5. 6. 7. 8.
I eat this food as a treat. I think this food is healthy. I have eaten this food (or a similar food) until I felt full. I have eaten this food (or a similar food) until I felt uncomfortably full/sick. I eat this food as a snack. I eat this food as a meal. I eat this food as part of a meal. I like this food.
15
M.A. Irvine et al. / Appetite 61 (2013) 13–18 Table 2 Attitudes toward, and familiarity with, the seven foods (data are mean +/ SE). Statement
Rank 1
Rank 2
Rank 3
Rank 4
Rank 5
Rank 6
Rank 7 (least)
Like food
Pasta 6.24 (0.16)
Pizza 5.98 (0.20)
Dairy milk 5.69 (0.25)
Banana 5.27 (0.30)
Crunchie 4.76 (0.29)
M&Ms. 4.76 (0.28)
Wine gums 4.20 (0.29)
Think is healthy
Banana 6.33 (0.20)
Pasta 5.51 (0.20)
Pizza 3.31 (0.18)
Dairy milk 2.24 (0.23)
M&Ms. 1.96 (0.15)
Wine gums 1.84 (0.15)
Crunchie 2.58 (0.22)
Eat if hungry
Pasta 5.31 (0.25)
Banana 4.18 (0.28)
Pizza 4.07 (0.27)
Dairy milk 2.93 (0.27)
Crunchie 2.58 (0.22)
M&Ms. 2.16 (0.18)
Wine gums 1.87 (0.20)
Eat as treat
Dairy milk 5.22 (0.24)
Crunchie 4.64 (0.27)
M&Ms. 4.62 (0.28)
Pizza 4.53 (0.24)
Wine gums 3.67 (0.29)
Pasta 2.93 (0.27)
Banana 2.53 (0.18)
Eat as part of meal
Pasta 5.53 (0.28)
Pizza 4.91 (0.29)
Banana 4.33 (0.26)
Dairy milk 2.33 (0.22)
Crunchie 1.91 (0.16)
Wine gums 1.62 (0.18)
M&Ms. 1.56 (0.12)
Eat as meal
Pasta 6.53 (0.15)
Pizza 5.64 (0.27)
Banana 2.27 (0.21)
Dairy milk 1.67 (0.21)
Crunchie 1.38 (0.11)
Wine gums 1.24 (0.14)
M&Ms. 1.20 (0.07)
Eat as snack
Banana 5.27 (0.29)
Dairy milk 4.44 (0.28)
M&Ms. 4.31 (0.31)
Crunchie 4.18 (0.31)
Wine gums 3.80 (0.30)
Pasta 3.13 (0.25)
Pizza 2.91 (0.26)
Can be part of balanced diet
Banana 6.62 (0.15)
Pasta 6.13 (0.19)
Pizza 4.53 (0.26)
Dairy milk 3.87 (0.30)
Crunchie 3.20 (0.25)
M&Ms. 3.18 (0.27)
Wine gums 2.64 (0.26)
Have eaten until full
Pasta 5.69 (0.21)
Pizza 5.02 (0.22)
Dairy milk 3.47 (0.26)
Banana 3.16 (0.23)
Crunchie 2.62 (0.21)
M&Ms. 2.62 (0.21)
Wine gums 2.29 (0.21)
Have eaten until overfull
Pizza 3.78 (0.27)
Dairy milk 3.44 (0.27)
Pasta 3.42 (0.27)
Crunchie 2.56 (0.24)
M&Ms. 2.29 (0.18)
Wine gums 2.20 (0.20)
Banana 1.82 (0.15)
Frequency of eat
Pasta 5.07 (0.14)
Banana 5.00 (0.28)
Dairy milk 4.07 (0.22)
Pizza 3.69 (0.14)
Crunchie 2.58 (0.16)
Wine gums 2.47 (0.17)
M&Ms. 2.36 (0.18)
Scoring key: see ‘‘Measures – Attitudes to Foods Task’’ section for Experiment 1.
9. I think this food can be part of a balanced diet. 10. I primarily eat this food if I am hungry. 11. I eat this food. For each statement participants selected an answer according to a 7-point scale. Statements 1 and 2, and 5–10, were anchored ‘‘very strongly disagree’’, and ‘‘neither agree nor disagree’’ (mid-point), to ‘‘very strongly agree’’. As these statements were not concerned with familiarity, the scale did not capture the frequency with which the foods were eaten. However statements 3 and 4 did pertain specifically to familiarity (of eating to fullness). Therefore the scale was necessarily different, with an explicit term attached to each point. These were: (1) never (2) very rarely (3) rarely (4) occasionally (5) quite often (6) often (7) very often. Statement 11 was the other measure specifically concerned with familiarity, this time of eating a food per se. Thus it also had a scale on which each point had a term attached, as follows: (1) never (2) less than once a month (3) about once a month (4) about once every 2 weeks (5) about once a week (6) more than once a week (7) almost every day. Both the order in which the statements, and the order in which the foods that the statements pertained to were presented, was randomised across participants. In order to run the task E-Prime (version 1.1; E-Studio 1.1, Copyright Psychology Software Tools, 1996–2002) was used. All questions were presented in Arial font, point size 24. Participants sat approximately 60 cm from the screen. Procedure After providing written consent participants were given an information sheet explaining the nature of the procedure. Participants were tested between 9 am and 5 pm on weekdays. Participants completed the psychophysical task first and then the Attitudes to Foods task. The 11 attitudinal and experiential statements appeared on the VDU for each of the seven foods in turn, with each statement being replaced with the next one contingent upon the participant pressing a number key from 1 to 7, in accordance with where they wanted to indicate their attitude on the 7-point scale.
At this juncture participants also indicated their height, age, and weight. There was no time limit stipulated for participants to complete any part of the test battery in. Overall, the test session took approximately 45 min. Upon completion, participants were given a debriefing form, which gave them background to the study, including the hypothesis and information drawn from relevant literature. Data analysis For each comparison-standard pair we used probit analysis to generate a point of subjective equality (PSE), which denotes the point at which two foods are expected to confer equal satiety. Each PSE was converted into a ratio by dividing the PSE by the energy content of the standard (255 or 260 kcal). Accordingly, a ratio of 1.3 would indicate that a particular comparison food is expected to be 1.3 times more filling than the standard (either wine gums or milk chocolate). Ratio values were then converted into Log10 units before being submitted to parametric analysis. Relationships between expected satiety and participants’ responses for the 11 attitudinal and experiential statements were assessed using Pearson’s product moment correlation coefficients. Two sets of PSEs were used, one when milk chocolate was the standard food and one when wine gums was the standard food. These scores were entered into the correlation matrix in conjunction with mean ratio scores for the 11 attitudinal/experiential statements. Mean ratio scores were obtained as follows: for ‘‘I like this food’’, for example, participants’ scores for milk chocolate were divided by the mean of their scores across the five comparison foods. A score above 1 indicates that an individual likes milk chocolate relatively more than the five comparison foods, whereas a score below 1 indicates the reverse. This procedure was repeated for wine gums, thereby generating the two sets of mean ratio scores required. Therefore, as for expected satiety, these ratios measure responses to chocolate and to wine gums relative to the other foods. On a few occasions (7.2%) the APE routine was unable to calculate an appropriate PSE, owing to its inability in these instances to select appropriate comparison food pictures around which the
16
M.A. Irvine et al. / Appetite 61 (2013) 13–18
standard food was selected 50% of the time. This, as Brunstrom et al. (2008) point out, is primarily attributable to the fact that sometimes the range of comparison pictures is too small. In such cases the PSE was treated as missing data. Results The rank order of foods for each of the statements in the attitude to foods task is shown in Table 2. In the main staple foods such as pasta and sauce and pizza are eaten to fullness more frequently than are snack foods. These types of foods also tend to be more liked. Table 3 displays the correlations, for milk chocolate and wine gums, between expected satiety (denoted by participants’ mean PSE scores) and the 11 attitudinal and experiential statements. For both foods there was a significant relationship between frequency of having eaten the food to fullness and PSE score, such that the food was estimated to be relatively less satiating if it had been eaten to fullness relatively less often. Interim discussion Experiment 1 sought to explore the relationship between expected satiety and familiarity. Of the 11 attitudinal/experiential measures only familiarity of eating to fullness was significantly correlated with expected satiety (PSE score) for both wine gums and milk chocolate. Other measures, such as liking, were not significant predictors of expected satiety. This result points toward a particular kind of familiarity, namely of eating to satiety, as being the most closely associated with expected satiety. Brunstrom et al. (2008) speculated that snack foods may be expected to confer less satiety because they tend not to be eaten to satiety, thereby diminishing the opportunity for learned satiety to occur. If learned satiety does indeed inform expected satiety then this could help to explain the present results. Furthermore according to Zandstra, Stubenitsky, deGraaf, and Mela (2002), and because there may not be sufficient post-ingestive feedback within a meal to terminate eating behaviour, conditioned satiety provides a means through which portion size can be informed by prior knowledge of the likely effects of eating a food. In this respect the similarities between learned satiety and expected satiety are obvious. The notion that there is a learned basis to satiety expectations may be best evidenced by demonstrating a change in expected satiety within a food, in accordance with a manipulation of participants’ level of familiarity with that food. Experiment 2 offers the opportunity for such a test, where any increases in expected satiety engendered by an eating episode will presumably be attributable to changes in degree of familiarity.
Table 3 Correlations between mean PSE scores and attitudinal/experiential measures. Mean PSE Dairy milk Eat frequency Eat as a meal Eat as a snack Eat as a treat Eat as part of a meal Eat until full Eat until sick Like food Eat if hungry Eat as part of balanced diet Think is healthy **
0.29 0.23 0.17 0.00 0.18 0.51** 0.29 0.25 0.23 0.06 0.17
Correlation is significant at the 0.01 level (2-tailed), df = 41.
Wine gums 0.17 0.22 0.24 0.08 0.08 0.40** 0.03 0.14 0.14 0.13 0.14
Method – Experiment 2 Participants The participants were 16 males and 28 females. They were mostly drawn from the undergraduate population of the University of Bristol. Their mean age was 19.4 [S.D. = 7.5] years, ranging from 18–59 years. No specific exclusion criteria were applied. The University of Bristol Faculty of Science Human Research Ethics Committee gave approval for the protocol. All participants were offered £7 sterling in remuneration for their assistance with the study. Measures Procedure The test battery took place over 2 days. Participants were firstly given an information sheet explaining the nature of the tasks (subsequent to providing written consent), and were tested between 9am and 5pm on weekdays, having firstly been asked to refrain from eating for 3 h before arriving for the first test session. This was because half of the participants would be (randomly) assigned to an ‘‘eat wine gums’’ condition, where they would be required to eat sufficient wine gums to confer a meaningful degree of purely wine gums-induced satiety. Participants firstly provided a measure of their hunger, and also details of when they last ate. The scales for these two questions were 7-point, and were as follows: For ‘‘When did you last eat?’’: (1) more than 12 h ago (2) 9–12 h ago (3) 6–9 h ago (4) 4–6 h ago (5) 2–4 h ago (6) 1–2 h ago (7) 0–1 h ago. For ‘‘How hungry are you right now?’’: (1) extremely full (2) very full (3) moderately full (4) neither hungry nor not hungry (5) moderately hungry (6) very hungry (7) extremely hungry. Following this participants undertook the psychophysical task and the attitudes to foods task. Those participants who were assigned to the wine gum-eating condition were then given the following instructions: ‘‘Please eat as many wine gums as necessary in order that you feel a level of fullness akin to that which you would experience after a typical meal (lunch or dinner). To reiterate, a comfortable level of fullness is what you should be aiming for. You will be given 15 min to eat the wine gums. More time will be made available should you require it. Please try to eat at least 25 wine gums’’ This bottom limit of 25 wine gums (500 kcals) was encouraged because such an amount should ensure a meaningful degree of satiety. After the eating episode these participants again indicated their level of hunger, and also their level of nausea, according to the following 7-point scale: ‘‘How sick are you right now?’’: (1) feel I am about to be sick (2) extremely sick (3) very sick (4) moderately sick (5) somewhat sick (6) a little sick (7) not at all sick In test session 2, which occurred between 1 and 4 days after test session 1, participants firstly indicated their levels of hunger and time since they last ate, and then completed the Psychophysical Task again. They were then given a debriefing form, which provided background information for the study. No time limit was stipulated for participants to complete the tasks, in either test session, and reaction times were not recorded. Test session 1 was approximately 70 min in duration; session 2 took approximately 50 min.
M.A. Irvine et al. / Appetite 61 (2013) 13–18
Fig. 1. Mean ratios (in Log10 units) of expected satiety for wine gums (upper panel) and milk chocolate (lower panel). A lower PSE indicates that the food is expected to be relatively more filling.
Data analysis PSEs were generated in an identical manner as described for Experiment 1. ANOVA was used to ascertain whether there were baseline differences between wine gum-eaters and non-eaters for any of the 11 attitudinal/experiential questions as they pertain to wine gums. Differences between groups’ mean PSE scores on test day 1 versus test day 2 were also assessed via repeated measures ANOVA: test day was the within-subjects factor. The Greenhouse-Geisser correction was applied as appropriate. As in Experiment 1, the APE routine was occasionally (8% of trials) unable to calculate an appropriate PSE. In such cases the PSE was again treated as missing data. Results The rank order for the attitudinal/experiential measures across foods was broadly very similar to that seen for Experiment 1 (see Table 2). In the interests of space a full table will not be reproduced for the present experiment. There was a significant test day wine gum group interaction for expected satiety for wine gums (F[1, 40] = 6.86, p < 0.05); but not for expected satiety for milk chocolate (F[1, 42] = 1.87, p = 0.18). Figure 1 shows that wine gum-eating participants expected wine gums to be relatively more filling on test day 2 versus test day 1 in comparison to participants who did not eat Wine gums (upper panel). By contrast expected satiety for milk chocolate did not change significantly for either group across test sessions (lower panel). Discussion Experiment 2 sought to establish whether an increase in familiarity (engendered by an eating episode) leads to a shift in expected satiety. The results of the expected satiety group analysis show this to be the case. For those participants who ate
17
wine gums to a substantial level of self-reported fullness, significant increases in expected satiety were seen after the eating episode: this was not the case for participants who did not eat wine gums. In the context of a learned acquisition hypothesis for expected satiety, these results are encouraging. The present results also demonstrate one-trial learning. It appears that a single experience of eating a food to satiety which is not habitually eaten in such a manner is enough to precipitate a significant reappraisal of satiety expectations for that food. This raises the issue of sufficient novelty: perhaps this result would not have been achieved if a food habitually eaten to satiety had been employed. The possibility of a general recency effect must also be considered, where perhaps eating any food to satiety would result in it subsequently being rated as more filling, regardless of a priori level of familiarity. A future experiment might address this by employing an additional, already-familiar food. If change in degree of familiarity is indeed driving the upwards-shift in expected satiety for wine gums observed herein, it would be expected that such a marked shift would not occur in the already-familiar food. There is in fact one other recent account of one-trial learning that is instructive to mention here, albeit in the context of expected satiation. Wilkinson and Brunstrom (2009) found that expected satiation for a novel high-energy-dense dessert (568 kcal) also increased significantly after a single eating episode. However it is interesting to note that these authors did not also find such an increase for a novel low-energy-dense dessert (228 kcal). This might reflect the fact that, of the two desserts, only the high-energydense version contained enough calories to confer satiety. This would tend to support the argument made presently, namely that eating a food to satiety is especially important in informing expected satiety beliefs. Future experiments might look to expand upon our methodology by assessing the impact of multiple eating episodes on expected satiety ratings, and also by establishing whether changes in expected satiety in response to eating episodes are maintained over longer time courses. Conclusion We have shown that expected satiety beliefs are associated with familiarity, in particular of having previously eaten a food to satiety. Furthermore, we find support for the notion that such beliefs have a learned basis, where increased familiarity of eating a relatively novel food to satiety precipitates significant increases in expected satiety.
References Brunstrom, J. M., Collingwood, J. M., & Rogers, P. J. (2010). Perceived volume, expected satiation, and the energy content of self-selected meals. Appetite, 55(1), 25–29. Brunstrom, J. M., & Shakeshaft, N. G. (2008). Measuring affective (liking) and nonaffective (expected satiety) determinants of portion size and food reward. Appetite, 52(1), 108–114. Brunstrom, J. M., Shakeshaft, N. G., & Scott-Samuel, N. E. (2008). Measuring ‘expected satiety’ in a range of common foods using a method of constant stimuli. Appetite, 51(3), 604–614. Holt, S. H., Brand-Miller, J. C., & Petocz, P. (1995). A satiety index of common foods. European Journal of Clinical Nutrition, 49, 675–690. Jeffery, R. W., Rydell, S., Dunn, C. L., Harnack, L. J., Levine, A. S., Pentel, P. R., et al. (2007). Effects of portion size on chronic energy intake. The International Journal of Behavioural Nutrition and Physical Activity, 4(27), 1–5. Krassner, H. A., Brownell, K. D., & Stunkard, A. J. (1979). Cleaning the plate. Food left over by overweight and normal weight persons. Behaviour Research and Therapy, 17(2), 155–156. Lebow, M. D., Chipperfield, J. G., & Magnusson, J. (1985). Leftovers, body-weight and sex of eater. Behaviour Research and Therapy, 23(2), 217. Rogers, P. J., & Smit, H. J. (2000). Food craving and ‘‘addiction’’. A critical review of the evidence from a biopsychosocial perspective. Pharmacology, Biochemistry & Behavior, 66(1), 3–14.
18
M.A. Irvine et al. / Appetite 61 (2013) 13–18
Rolls, B. J., Roe, L. S., Kral, T. V., Meengs, J. S., & Wall, D. E. (2004a). Increasing the portion size of a packaged snack increases energy intake in men and women. Appetite, 42(1), 63–69. Rolls, B. J., Roe, L. S., & Meengs, J. S. (2006a). Larger portion sizes lead to a sustained increase in energy intake over 2 days. Journal of the American Dietetic Association, 106(4), 543–549. Rolls, B. J., Roe, L. S., & Meengs, J. S. (2006b). Reductions in portion size and energy density of foods are additive and lead to sustained decreases in energy intake. American Journal of Clinical Nutrition, 83(1), 11–17. Rolls, B. J., Roe, L. S., & Meengs, J. S. (2007). The effect of large portion sizes on energy intake is sustained for 11 days. Obesity, 15(6), 1535–1543. Rolls, B. J., Roe, L. S., Meengs, J. S., & Wall, D. E. (2004b). Increasing the portion size of a sandwich increases energy intake. Journal of the American Dietetic Association, 104(3), 367–372.
Wansink, B., & Cheney, M. M. (2005). Super bowls. Serving bowl size and food consumption. Journal of the American Medical Association, 293(14), 1727–1728. Wansink, B., Painter, J. E., & North, J. (2005). Bottomless bowls. Why visual cues of portion size may influence intake. Obesity Research, 13, 93–100. Wilkinson, L. L., & Brunstrom, J. M. (2009). Conditioning ‘fullness expectations’ in a novel dessert. Appetite, 52, 780–783. Zandstra, E. H., Stubenitsky, K., deGraaf, C., & Mela, D. J. (2002). Effects of learned flavour cues on short-term regulation of food intake in a realistic setting. Physiology & Behavior, 75, 83–90.
Contents lists available at SciVerse ScienceDirect
Appetite journal homepage: www.elsevier.com/locate/appet
Research report
Increased familiarity with eating a food to fullness underlies increased expected satiety q Michael A. Irvine a,c,⇑, Jeffrey M. Brunstrom a, Philip Gee b, Peter J. Rogers a a
School of Experimental Psychology, University of Bristol, 12a Priory Road, Bristol BS8 1TU, England, United Kingdom School of Psychology, University of Plymouth, Drake Circus, Plymouth PL4 8AA, England, United Kingdom c Department of Psychiatry, University of Cambridge, Addenbrooke’s Hospital site, Level E4, P.O. Box 189, Cambridge CB2 0QQ, United Kingdom b
a r t i c l e
i n f o
Article history: Received 29 March 2012 Received in revised form 19 September 2012 Accepted 11 October 2012 Available online 22 October 2012 Keywords: Expected satiety Expected satiation Palatability Energy density Associative learning Flavour–nutrient learning Learned satiety
a b s t r a c t Expected satiety informs self-selected portion sizes and thereby influences energy intake. At present the extent to which these beliefs are learned remains unclear. In an initial study the proposition that familiarity influences expected satiety was explored. Self-report measures of familiarity, along with other measures such as degree of liking, were collected for wine gums and milk chocolate, together with expected satiety estimates obtained using a psychophysical task. Familiarity was indeed significantly correlated with expected satiety, but only in respect of frequency of having eaten the food to fullness. In a second experiment a significant increase in expected satiety was observed after eating a large portion of wine gums at a subsequent test session. Together, these findings indicate that expected satiety changes in response to increased familiarity of eating a food to satiety. Ó 2012 Elsevier Ltd. All rights reserved.
Introduction Prospective judgements relating to the ‘‘fillingness’’ of foods are referred to as decisions based on ‘‘expected satiety’’. Somewhat surprisingly, given their potential import to energy intake, these judgements have historically been given little consideration. As Brunstrom and Shakeshaft (2008) report, however, satiety expectations are an important determinant of self-selected portion size (kcal), and as such they would appear to be very relevant to the area of weight control. This is because portion size has been shown to be an excellent predictor of amount of food consumed (Jeffery et al., 2007; Rolls, Roe, Kral, Meengs, & Wall, 2004a; Rolls, Roe, & Meengs, 2006a, 2006b, 2007; Rolls, Roe, Meengs, & Wall, 2004b; Wansink, Painter, & North, 2005), where self-selected portions are often eaten in their entirety (Krassner, Brownell, & Stunkard, 1979; Lebow, Chipperfield, & Magnusson, 1985; Wansink & Cheney, 2005). Indeed, and even after controlling for volume, Brunstrom, Collingwood, and Rogers (2010) found a close correspondence between expected satiation and self-selected portion size.
q Acknowledgments: This work was supported by Cadbury plc, a member of Kraft FoodsTM and Great Western Research. A preliminary report of these experiments was presented at the 8th Sensory Science Symposium, Florence, Italy (2009). ⇑ Corresponding author. E-mail address: [email protected] (M.A. Irvine).
0195-6663/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.appet.2012.10.011
Presumably, eating experiences shape beliefs about the expected ‘‘fillingness’’ of foods. Therefore expected satiety should correspond with the actual degree of satiety occurring after ingestion. Supporting this, Brunstrom, Shakeshaft, and Scott-Samuel (2008) reports a highly significant correlation between the expected satiety scores of participants in their study with the actual satiety scores derived in a study by Holt, Brand-Miller, and Petocz (1995). In addition these authors found that familiar foods are expected to be more satiating. One possibility is that different aspects of familiarity – for example familiarity with eating a food per se (that is, in any caloric denomination) versus familiarity of eating it in sufficient quantities so as to confer satiety – might be differentially important in informing satiety expectations. Brunstrom et al. (2008) suggests that certain foods may be expected to confer relatively less satiety (kcal for kcal) because they are not often eaten to satiety, thereby limiting the opportunity for learned satiety to occur. Furthermore, Rogers and Smit (2000) allude to the possibility that snack foods may be conceptualised as ‘‘moreish’’ due to the fact societal norms dictate that such foods are relatively rarely eaten in amounts equivalent to staple foods (i.e., in amounts that generate significant to strong feelings of fullness). In light of the above discussion, the aim of the present paper was to further explore the relationship between expected satiety and familiarity, and in particular familiarity of eating to fullness. In Experiment 1 we sought to establish whether expected satiety
14
M.A. Irvine et al. / Appetite 61 (2013) 13–18
is correlated with self-reported ratings of familiarity for milk chocolate and wine gums (a chewy, pastille-type sweet/candy, which is generally eaten less frequently than chocolate). Familiarity was indexed by dissociable measures specifically pertaining to either familiarity of eating a food per se, or familiarity of eating it to fullness. In Experiment 2 an eating component was introduced: ratings of expected satiety for wine gums (the relatively novel food) were recorded before and after an eating episode. We reasoned that a difference between pre- and post-eating episode ratings would provide evidence of a direct influence of (increased) familiarity on expected satiety beliefs. Method – Experiment 1 Participants The participants were 7 males and 39 females. They were drawn from the undergraduate population of the University of Bristol. Their mean age was 19.4 [S.D. = 5.8] years. No specific exclusion criteria were applied. They participated as part of their course requirement, or were offered £7 Sterling in remuneration for their assistance with the study. Most participants were of normal body weight. The University of Bristol Faculty of Science Human Research Ethics Committee gave approval for the protocol. Design Images of seven different foods were presented in 260-kilocalorie (kcal) portions, as follows: pasta and sauce (‘‘Egg Penne’’ pasta, Sainsbury’s UK, Sainsbury’s Supermarkets Ltd., Holborn, London, UK); ‘‘Sun-dried stir in tomato sauce’’ (Dolmio, Masterfoods, Melton Mowbray, Leicestershire, UK); pizza (‘‘Deeply Delicious Margherita’’, Goodfellas, Northern Foods Frozen, Salford Quays, Manchester, UK); Dairy Milk, Crunchie, and Maynard’s Wine Gums (all three by Cadbury plc, Bournville, Birmingham, UK); Peanut M&Ms (Mars, Mars UK Ltd., Slough, Berkshire, UK); and banana. Dairy Milk is a solid milk chocolate bar; Crunchie is a honeycomb-centred, chocolate-covered bar; and Wine Gums are chewy, firm, fruit-flavoured pastille-type sweets/candy. These foods were used because previous studies have shown that they vary in familiarity, liking, and the typical context in which they are eaten (Irvine et al., unpublished data). For the present study data pertaining to these measures are given in Table 2 (see Results section). The macronutrient composition and energy density of these foods are shown in Table 1. Measures Psychophysical task An adapted version of the psychophysical methodology first introduced by Brunstrom et al. (2008) was employed to assess expected satiety. This methodology is based on a ‘‘method of con-
Table 1 Macronutrient composition and energy density of the seven foods. Values for carbohydrate, protein, fat, fibre and kcals are per 100 g. Foods are ordered from least to most energy dense. Food
Carbohydrate
Protein
Fat
Fibre
kcal
Banana Pasta & sauce Wine gums Pizza Crunchie M&Ms Dairy milk
23.2 21.4 75 40.2 69.9 46.4 56.7
1.2 5.3 6.1 16.5 3.6 9.6 7.5
0.3 5.3 0.2 13.2 19.1 26.2 29.8
1.1 3.1 0 4 1.3 2.8 0.7
95 153 325 346 465 515 525
stant stimuli’’ (MOCS). The MOCS is well suited to the measurement of expected satiety across different foods. In comparison to methods where estimates are based on a single decision, the MOCS is used to derive estimates based upon many decisions, which promotes greater accuracy. Estimates of expected satiety were obtained for the same seven foods described above. Five of the foods were designated ‘‘comparison foods’’. The other two, milk chocolate and wine gums, were the ‘‘standard foods’’, and were always depicted in 255 kcal (milk chocolate) and 260 kcal (wine gums) portions. These caloric denominations were used because there are 255 kcals in a standard 49 g bar of Dairy Milk™, and there are 260 kcals in 13 wine gums, which gives the closest match to the Dairy Milk bar without having to cut up any wine gums. The assignment of a food as a comparison or a standard is somewhat arbitrary, in that the MOCS procedure can work perfectly well in any configuration (see for example Brunstrom et al., 2008). As such there is no basis on which to predict that results will depend upon which foods were designated to which category. At any given time the computer screen displayed a photograph of one of the two standard foods on the left and one of the five comparison foods on the right, arranged on a 255-mm diameter white plate. The images were of a high quality, and were taken using a camera mounted directly overhead. For each comparison food there were between 40 and 70 photographs taken, of different portion sizes varying in 20 kcal increments. As such, the maximum displayed portion sizes for the comparison foods varied between 800 kcal and 1400 kcal. This difference was due to the fact some of the less energy-dense foods could not be accommodated on the plate in very large portions (by weight). During the task the participants pressed the left cursor key if they thought the food portion on the left would make them feel the most full, or the right cursor key to indicate the portion on the right. After each key press the next two portions of food would appear. On each trial the amount and type of comparison food changed. Each standard food was paired with each comparison food, in various caloric denominations, on multiple occasions. After a sufficient number of trials the probability that the standard will be selected over the range of comparison values can be plotted: probit analysis can be used to fit a sigmoid function, from which a point of subjective equality (PSE) can be extrapolated. As noted by Brunstrom et al. (2008), this value is important because it ‘‘indicates the amount of the comparison (kcal) that is expected to be equally as filling as the standard’’. Thus, and by making a set of systematic comparisons between a common standard/s and a range of comparison foods, precise expected satiety scores can be derived and compared across a range of foods. In total participants responded to 540 trials (54 pairings of each of the five comparison foods with each of the two standard foods). Attitudes to foods task In this task participants responded to 11 different statements, which were to be correlated with estimates of expected satiety. To reiterate, the rationale was to assess the relationship between expected satiety and attitudes towards/level of familiarity with the foods. The statements were as follows: 1. 2. 3. 4. 5. 6. 7. 8.
I eat this food as a treat. I think this food is healthy. I have eaten this food (or a similar food) until I felt full. I have eaten this food (or a similar food) until I felt uncomfortably full/sick. I eat this food as a snack. I eat this food as a meal. I eat this food as part of a meal. I like this food.
15
M.A. Irvine et al. / Appetite 61 (2013) 13–18 Table 2 Attitudes toward, and familiarity with, the seven foods (data are mean +/ SE). Statement
Rank 1
Rank 2
Rank 3
Rank 4
Rank 5
Rank 6
Rank 7 (least)
Like food
Pasta 6.24 (0.16)
Pizza 5.98 (0.20)
Dairy milk 5.69 (0.25)
Banana 5.27 (0.30)
Crunchie 4.76 (0.29)
M&Ms. 4.76 (0.28)
Wine gums 4.20 (0.29)
Think is healthy
Banana 6.33 (0.20)
Pasta 5.51 (0.20)
Pizza 3.31 (0.18)
Dairy milk 2.24 (0.23)
M&Ms. 1.96 (0.15)
Wine gums 1.84 (0.15)
Crunchie 2.58 (0.22)
Eat if hungry
Pasta 5.31 (0.25)
Banana 4.18 (0.28)
Pizza 4.07 (0.27)
Dairy milk 2.93 (0.27)
Crunchie 2.58 (0.22)
M&Ms. 2.16 (0.18)
Wine gums 1.87 (0.20)
Eat as treat
Dairy milk 5.22 (0.24)
Crunchie 4.64 (0.27)
M&Ms. 4.62 (0.28)
Pizza 4.53 (0.24)
Wine gums 3.67 (0.29)
Pasta 2.93 (0.27)
Banana 2.53 (0.18)
Eat as part of meal
Pasta 5.53 (0.28)
Pizza 4.91 (0.29)
Banana 4.33 (0.26)
Dairy milk 2.33 (0.22)
Crunchie 1.91 (0.16)
Wine gums 1.62 (0.18)
M&Ms. 1.56 (0.12)
Eat as meal
Pasta 6.53 (0.15)
Pizza 5.64 (0.27)
Banana 2.27 (0.21)
Dairy milk 1.67 (0.21)
Crunchie 1.38 (0.11)
Wine gums 1.24 (0.14)
M&Ms. 1.20 (0.07)
Eat as snack
Banana 5.27 (0.29)
Dairy milk 4.44 (0.28)
M&Ms. 4.31 (0.31)
Crunchie 4.18 (0.31)
Wine gums 3.80 (0.30)
Pasta 3.13 (0.25)
Pizza 2.91 (0.26)
Can be part of balanced diet
Banana 6.62 (0.15)
Pasta 6.13 (0.19)
Pizza 4.53 (0.26)
Dairy milk 3.87 (0.30)
Crunchie 3.20 (0.25)
M&Ms. 3.18 (0.27)
Wine gums 2.64 (0.26)
Have eaten until full
Pasta 5.69 (0.21)
Pizza 5.02 (0.22)
Dairy milk 3.47 (0.26)
Banana 3.16 (0.23)
Crunchie 2.62 (0.21)
M&Ms. 2.62 (0.21)
Wine gums 2.29 (0.21)
Have eaten until overfull
Pizza 3.78 (0.27)
Dairy milk 3.44 (0.27)
Pasta 3.42 (0.27)
Crunchie 2.56 (0.24)
M&Ms. 2.29 (0.18)
Wine gums 2.20 (0.20)
Banana 1.82 (0.15)
Frequency of eat
Pasta 5.07 (0.14)
Banana 5.00 (0.28)
Dairy milk 4.07 (0.22)
Pizza 3.69 (0.14)
Crunchie 2.58 (0.16)
Wine gums 2.47 (0.17)
M&Ms. 2.36 (0.18)
Scoring key: see ‘‘Measures – Attitudes to Foods Task’’ section for Experiment 1.
9. I think this food can be part of a balanced diet. 10. I primarily eat this food if I am hungry. 11. I eat this food. For each statement participants selected an answer according to a 7-point scale. Statements 1 and 2, and 5–10, were anchored ‘‘very strongly disagree’’, and ‘‘neither agree nor disagree’’ (mid-point), to ‘‘very strongly agree’’. As these statements were not concerned with familiarity, the scale did not capture the frequency with which the foods were eaten. However statements 3 and 4 did pertain specifically to familiarity (of eating to fullness). Therefore the scale was necessarily different, with an explicit term attached to each point. These were: (1) never (2) very rarely (3) rarely (4) occasionally (5) quite often (6) often (7) very often. Statement 11 was the other measure specifically concerned with familiarity, this time of eating a food per se. Thus it also had a scale on which each point had a term attached, as follows: (1) never (2) less than once a month (3) about once a month (4) about once every 2 weeks (5) about once a week (6) more than once a week (7) almost every day. Both the order in which the statements, and the order in which the foods that the statements pertained to were presented, was randomised across participants. In order to run the task E-Prime (version 1.1; E-Studio 1.1, Copyright Psychology Software Tools, 1996–2002) was used. All questions were presented in Arial font, point size 24. Participants sat approximately 60 cm from the screen. Procedure After providing written consent participants were given an information sheet explaining the nature of the procedure. Participants were tested between 9 am and 5 pm on weekdays. Participants completed the psychophysical task first and then the Attitudes to Foods task. The 11 attitudinal and experiential statements appeared on the VDU for each of the seven foods in turn, with each statement being replaced with the next one contingent upon the participant pressing a number key from 1 to 7, in accordance with where they wanted to indicate their attitude on the 7-point scale.
At this juncture participants also indicated their height, age, and weight. There was no time limit stipulated for participants to complete any part of the test battery in. Overall, the test session took approximately 45 min. Upon completion, participants were given a debriefing form, which gave them background to the study, including the hypothesis and information drawn from relevant literature. Data analysis For each comparison-standard pair we used probit analysis to generate a point of subjective equality (PSE), which denotes the point at which two foods are expected to confer equal satiety. Each PSE was converted into a ratio by dividing the PSE by the energy content of the standard (255 or 260 kcal). Accordingly, a ratio of 1.3 would indicate that a particular comparison food is expected to be 1.3 times more filling than the standard (either wine gums or milk chocolate). Ratio values were then converted into Log10 units before being submitted to parametric analysis. Relationships between expected satiety and participants’ responses for the 11 attitudinal and experiential statements were assessed using Pearson’s product moment correlation coefficients. Two sets of PSEs were used, one when milk chocolate was the standard food and one when wine gums was the standard food. These scores were entered into the correlation matrix in conjunction with mean ratio scores for the 11 attitudinal/experiential statements. Mean ratio scores were obtained as follows: for ‘‘I like this food’’, for example, participants’ scores for milk chocolate were divided by the mean of their scores across the five comparison foods. A score above 1 indicates that an individual likes milk chocolate relatively more than the five comparison foods, whereas a score below 1 indicates the reverse. This procedure was repeated for wine gums, thereby generating the two sets of mean ratio scores required. Therefore, as for expected satiety, these ratios measure responses to chocolate and to wine gums relative to the other foods. On a few occasions (7.2%) the APE routine was unable to calculate an appropriate PSE, owing to its inability in these instances to select appropriate comparison food pictures around which the
16
M.A. Irvine et al. / Appetite 61 (2013) 13–18
standard food was selected 50% of the time. This, as Brunstrom et al. (2008) point out, is primarily attributable to the fact that sometimes the range of comparison pictures is too small. In such cases the PSE was treated as missing data. Results The rank order of foods for each of the statements in the attitude to foods task is shown in Table 2. In the main staple foods such as pasta and sauce and pizza are eaten to fullness more frequently than are snack foods. These types of foods also tend to be more liked. Table 3 displays the correlations, for milk chocolate and wine gums, between expected satiety (denoted by participants’ mean PSE scores) and the 11 attitudinal and experiential statements. For both foods there was a significant relationship between frequency of having eaten the food to fullness and PSE score, such that the food was estimated to be relatively less satiating if it had been eaten to fullness relatively less often. Interim discussion Experiment 1 sought to explore the relationship between expected satiety and familiarity. Of the 11 attitudinal/experiential measures only familiarity of eating to fullness was significantly correlated with expected satiety (PSE score) for both wine gums and milk chocolate. Other measures, such as liking, were not significant predictors of expected satiety. This result points toward a particular kind of familiarity, namely of eating to satiety, as being the most closely associated with expected satiety. Brunstrom et al. (2008) speculated that snack foods may be expected to confer less satiety because they tend not to be eaten to satiety, thereby diminishing the opportunity for learned satiety to occur. If learned satiety does indeed inform expected satiety then this could help to explain the present results. Furthermore according to Zandstra, Stubenitsky, deGraaf, and Mela (2002), and because there may not be sufficient post-ingestive feedback within a meal to terminate eating behaviour, conditioned satiety provides a means through which portion size can be informed by prior knowledge of the likely effects of eating a food. In this respect the similarities between learned satiety and expected satiety are obvious. The notion that there is a learned basis to satiety expectations may be best evidenced by demonstrating a change in expected satiety within a food, in accordance with a manipulation of participants’ level of familiarity with that food. Experiment 2 offers the opportunity for such a test, where any increases in expected satiety engendered by an eating episode will presumably be attributable to changes in degree of familiarity.
Table 3 Correlations between mean PSE scores and attitudinal/experiential measures. Mean PSE Dairy milk Eat frequency Eat as a meal Eat as a snack Eat as a treat Eat as part of a meal Eat until full Eat until sick Like food Eat if hungry Eat as part of balanced diet Think is healthy **
0.29 0.23 0.17 0.00 0.18 0.51** 0.29 0.25 0.23 0.06 0.17
Correlation is significant at the 0.01 level (2-tailed), df = 41.
Wine gums 0.17 0.22 0.24 0.08 0.08 0.40** 0.03 0.14 0.14 0.13 0.14
Method – Experiment 2 Participants The participants were 16 males and 28 females. They were mostly drawn from the undergraduate population of the University of Bristol. Their mean age was 19.4 [S.D. = 7.5] years, ranging from 18–59 years. No specific exclusion criteria were applied. The University of Bristol Faculty of Science Human Research Ethics Committee gave approval for the protocol. All participants were offered £7 sterling in remuneration for their assistance with the study. Measures Procedure The test battery took place over 2 days. Participants were firstly given an information sheet explaining the nature of the tasks (subsequent to providing written consent), and were tested between 9am and 5pm on weekdays, having firstly been asked to refrain from eating for 3 h before arriving for the first test session. This was because half of the participants would be (randomly) assigned to an ‘‘eat wine gums’’ condition, where they would be required to eat sufficient wine gums to confer a meaningful degree of purely wine gums-induced satiety. Participants firstly provided a measure of their hunger, and also details of when they last ate. The scales for these two questions were 7-point, and were as follows: For ‘‘When did you last eat?’’: (1) more than 12 h ago (2) 9–12 h ago (3) 6–9 h ago (4) 4–6 h ago (5) 2–4 h ago (6) 1–2 h ago (7) 0–1 h ago. For ‘‘How hungry are you right now?’’: (1) extremely full (2) very full (3) moderately full (4) neither hungry nor not hungry (5) moderately hungry (6) very hungry (7) extremely hungry. Following this participants undertook the psychophysical task and the attitudes to foods task. Those participants who were assigned to the wine gum-eating condition were then given the following instructions: ‘‘Please eat as many wine gums as necessary in order that you feel a level of fullness akin to that which you would experience after a typical meal (lunch or dinner). To reiterate, a comfortable level of fullness is what you should be aiming for. You will be given 15 min to eat the wine gums. More time will be made available should you require it. Please try to eat at least 25 wine gums’’ This bottom limit of 25 wine gums (500 kcals) was encouraged because such an amount should ensure a meaningful degree of satiety. After the eating episode these participants again indicated their level of hunger, and also their level of nausea, according to the following 7-point scale: ‘‘How sick are you right now?’’: (1) feel I am about to be sick (2) extremely sick (3) very sick (4) moderately sick (5) somewhat sick (6) a little sick (7) not at all sick In test session 2, which occurred between 1 and 4 days after test session 1, participants firstly indicated their levels of hunger and time since they last ate, and then completed the Psychophysical Task again. They were then given a debriefing form, which provided background information for the study. No time limit was stipulated for participants to complete the tasks, in either test session, and reaction times were not recorded. Test session 1 was approximately 70 min in duration; session 2 took approximately 50 min.
M.A. Irvine et al. / Appetite 61 (2013) 13–18
Fig. 1. Mean ratios (in Log10 units) of expected satiety for wine gums (upper panel) and milk chocolate (lower panel). A lower PSE indicates that the food is expected to be relatively more filling.
Data analysis PSEs were generated in an identical manner as described for Experiment 1. ANOVA was used to ascertain whether there were baseline differences between wine gum-eaters and non-eaters for any of the 11 attitudinal/experiential questions as they pertain to wine gums. Differences between groups’ mean PSE scores on test day 1 versus test day 2 were also assessed via repeated measures ANOVA: test day was the within-subjects factor. The Greenhouse-Geisser correction was applied as appropriate. As in Experiment 1, the APE routine was occasionally (8% of trials) unable to calculate an appropriate PSE. In such cases the PSE was again treated as missing data. Results The rank order for the attitudinal/experiential measures across foods was broadly very similar to that seen for Experiment 1 (see Table 2). In the interests of space a full table will not be reproduced for the present experiment. There was a significant test day wine gum group interaction for expected satiety for wine gums (F[1, 40] = 6.86, p < 0.05); but not for expected satiety for milk chocolate (F[1, 42] = 1.87, p = 0.18). Figure 1 shows that wine gum-eating participants expected wine gums to be relatively more filling on test day 2 versus test day 1 in comparison to participants who did not eat Wine gums (upper panel). By contrast expected satiety for milk chocolate did not change significantly for either group across test sessions (lower panel). Discussion Experiment 2 sought to establish whether an increase in familiarity (engendered by an eating episode) leads to a shift in expected satiety. The results of the expected satiety group analysis show this to be the case. For those participants who ate
17
wine gums to a substantial level of self-reported fullness, significant increases in expected satiety were seen after the eating episode: this was not the case for participants who did not eat wine gums. In the context of a learned acquisition hypothesis for expected satiety, these results are encouraging. The present results also demonstrate one-trial learning. It appears that a single experience of eating a food to satiety which is not habitually eaten in such a manner is enough to precipitate a significant reappraisal of satiety expectations for that food. This raises the issue of sufficient novelty: perhaps this result would not have been achieved if a food habitually eaten to satiety had been employed. The possibility of a general recency effect must also be considered, where perhaps eating any food to satiety would result in it subsequently being rated as more filling, regardless of a priori level of familiarity. A future experiment might address this by employing an additional, already-familiar food. If change in degree of familiarity is indeed driving the upwards-shift in expected satiety for wine gums observed herein, it would be expected that such a marked shift would not occur in the already-familiar food. There is in fact one other recent account of one-trial learning that is instructive to mention here, albeit in the context of expected satiation. Wilkinson and Brunstrom (2009) found that expected satiation for a novel high-energy-dense dessert (568 kcal) also increased significantly after a single eating episode. However it is interesting to note that these authors did not also find such an increase for a novel low-energy-dense dessert (228 kcal). This might reflect the fact that, of the two desserts, only the high-energydense version contained enough calories to confer satiety. This would tend to support the argument made presently, namely that eating a food to satiety is especially important in informing expected satiety beliefs. Future experiments might look to expand upon our methodology by assessing the impact of multiple eating episodes on expected satiety ratings, and also by establishing whether changes in expected satiety in response to eating episodes are maintained over longer time courses. Conclusion We have shown that expected satiety beliefs are associated with familiarity, in particular of having previously eaten a food to satiety. Furthermore, we find support for the notion that such beliefs have a learned basis, where increased familiarity of eating a relatively novel food to satiety precipitates significant increases in expected satiety.
References Brunstrom, J. M., Collingwood, J. M., & Rogers, P. J. (2010). Perceived volume, expected satiation, and the energy content of self-selected meals. Appetite, 55(1), 25–29. Brunstrom, J. M., & Shakeshaft, N. G. (2008). Measuring affective (liking) and nonaffective (expected satiety) determinants of portion size and food reward. Appetite, 52(1), 108–114. Brunstrom, J. M., Shakeshaft, N. G., & Scott-Samuel, N. E. (2008). Measuring ‘expected satiety’ in a range of common foods using a method of constant stimuli. Appetite, 51(3), 604–614. Holt, S. H., Brand-Miller, J. C., & Petocz, P. (1995). A satiety index of common foods. European Journal of Clinical Nutrition, 49, 675–690. Jeffery, R. W., Rydell, S., Dunn, C. L., Harnack, L. J., Levine, A. S., Pentel, P. R., et al. (2007). Effects of portion size on chronic energy intake. The International Journal of Behavioural Nutrition and Physical Activity, 4(27), 1–5. Krassner, H. A., Brownell, K. D., & Stunkard, A. J. (1979). Cleaning the plate. Food left over by overweight and normal weight persons. Behaviour Research and Therapy, 17(2), 155–156. Lebow, M. D., Chipperfield, J. G., & Magnusson, J. (1985). Leftovers, body-weight and sex of eater. Behaviour Research and Therapy, 23(2), 217. Rogers, P. J., & Smit, H. J. (2000). Food craving and ‘‘addiction’’. A critical review of the evidence from a biopsychosocial perspective. Pharmacology, Biochemistry & Behavior, 66(1), 3–14.
18
M.A. Irvine et al. / Appetite 61 (2013) 13–18
Rolls, B. J., Roe, L. S., Kral, T. V., Meengs, J. S., & Wall, D. E. (2004a). Increasing the portion size of a packaged snack increases energy intake in men and women. Appetite, 42(1), 63–69. Rolls, B. J., Roe, L. S., & Meengs, J. S. (2006a). Larger portion sizes lead to a sustained increase in energy intake over 2 days. Journal of the American Dietetic Association, 106(4), 543–549. Rolls, B. J., Roe, L. S., & Meengs, J. S. (2006b). Reductions in portion size and energy density of foods are additive and lead to sustained decreases in energy intake. American Journal of Clinical Nutrition, 83(1), 11–17. Rolls, B. J., Roe, L. S., & Meengs, J. S. (2007). The effect of large portion sizes on energy intake is sustained for 11 days. Obesity, 15(6), 1535–1543. Rolls, B. J., Roe, L. S., Meengs, J. S., & Wall, D. E. (2004b). Increasing the portion size of a sandwich increases energy intake. Journal of the American Dietetic Association, 104(3), 367–372.
Wansink, B., & Cheney, M. M. (2005). Super bowls. Serving bowl size and food consumption. Journal of the American Medical Association, 293(14), 1727–1728. Wansink, B., Painter, J. E., & North, J. (2005). Bottomless bowls. Why visual cues of portion size may influence intake. Obesity Research, 13, 93–100. Wilkinson, L. L., & Brunstrom, J. M. (2009). Conditioning ‘fullness expectations’ in a novel dessert. Appetite, 52, 780–783. Zandstra, E. H., Stubenitsky, K., deGraaf, C., & Mela, D. J. (2002). Effects of learned flavour cues on short-term regulation of food intake in a realistic setting. Physiology & Behavior, 75, 83–90.